Abstract

Computational intelligence has been frequently applied to modeling artificial agents in agent-based computational economics. Commonly used applications include reinforcement learning (Chen, 2013), classifier systems (Vriend, 2002), genetic algorithms, genetic programming (Chen, 2002a, b), swarm intelligence (Boyer, Brorsen & Zhang, 2014), and instance-based learning (Pape & Kurtz, 2013). They are considered as alternative toolkits for the classical or Bayesian statistical models in modeling bounded-rationality and adaptive behavior (Sargent, 1993). However, these toolkits, except for reinforcement learning, are not explicitly grounded in psychology. It, therefore, remains to be seen whether these “machines” (artificial agents) are related to the bounded-rational agents as conceived by behavioral economists. Or, alternatively, to what extent can we relate the general principles or practices that are frequently applied in behavioral economics to the designs of these machines?